A stereo technology is for the purpose of transmitting or reconstructing a certain specified sound field, so as to reproduce acoustic and spatial characteristics of an original sound field for listeners. In recent years, with the development of a computer technology and digital signal processing technology, and due to the needs of development of high-definition television sound systems and home audiovisual systems, the stereo technology has undergone significant development, and meanwhile, higher requirements are imposed on the stereo technology, especially stereo encoding and decoding technologies.
The conventional stereo encoding methods may be categorized into two types: one type is early waveform-based stereo encoding method, and the other type is currently commonly-used parametric stereo encoding method. In the parametric stereo encoding method, generally, left and right channel signals are down-mixed rather than being directly encoded, the down-mixed signals are encoded, and some extra sideband information is also encoded. At a decoding end, a stereo signal is recovered by using the down-mixed signals and the sideband information.
The quality of the stereo signal depends, to a large extent, on the quality of the down-mixed signals. The more synchronous are the left and right channel signals, the less information is lost in the down-mixing process. Generally, distances from a sound emitting object to two microphones recording sounds the left and right channels may change or be different, which inevitably leads to a delay between the left and right channel signals. The left and right channel signals cannot be completely synchronized. If the delay can be adjusted in the down-mixing process, that is, the left and right channel signals are synchronized, the quality of the synthesized stereo signal may be improved to a great extent.
FIG. 1 is a schematic flow chart of a stereo encoding method in the prior art. Referring to FIG. 1, firstly, a residual signal is obtained by performing down-sampling 4, Linear Predictive Coding (LPC) analysis, and LPC filtering on the left and right channel signals. Then, delays of the left and right channel signals are respectively extracted, and if the delays of two continuous frames of the left and right channel signals are different, a delay adjustment is performed before the down-mixing process.
In the process of implementing the present invention, the inventor finds that:
Because the left and right channel signals need to be spliced and added in the delay adjustment process, distortion is introduced, and the stereo signals with different characteristics have different distortion effects on discontinuity of interframe data during the splicing and adding process. According to the prior art, as the characteristics of the stereo signals are not differentiated during a delay adjustment, and the delay adjustment is performed immediately as long as delays of two continuous frames of the left and right channel signals are different, serious distortion may be caused.